I've been meaning to post this for a couple of weeks but never quite
decided what the point was, so I'll just give you a coredump :-)

This year, UTC acquired an HP3000 969/120 to replace
maintenance-expiring 3000/950 (Library) and 3000/960 (Administrative),
along with an HP9000/K570 to run the "new, integrated" Administrative
systems (Oracle-based garbage).

Last fall, during bulk fee payment/late registration/drop-adds, we
peaked the 960 at around 160 users and had 100% CPU utilization with run
queues getting into the double digits.  Response times were reaching
double-digit seconds.  The library's 950 was painfully slow but
tolerable as it was only serving 24 or so users.

Earlier this month we had spring bulk fee payment/late registration/etc
on the new 969/120.  We peaked at 208 users.  The first day we averaged
24% CPU utilization, the 2nd day only 18%.  Note that the library is on
here too now and was smoking with web search results
(http://www.library.utc.edu) being served from a sizeable Image
database.  Response times all around were incredibly amazing.

The 9000 has been in place almost a year.  We're still in
pre-implementation.  It's a 4-way processor with 2 gigs of RAM, dual
fiber channel Nike 30 w/144Gb, dwarfing the 969/120 by at least an order
of magnitude.  I spent the majority of last week installing TWO HUNDRED
AND FORTY TWO patches to HPUX and those just accumulated since last
August, and represent only those deemed important enough to be
mandatory.

What is wrong with this picture?

Jeff Kell <[log in to unmask]>